This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
utf8.h, utfebcdic.h: Add, fix comments
[perl5.git] / utf8.h
CommitLineData
a0ed51b3
LW
1/* utf8.h
2 *
f3cb6f94
KW
3 * This file contains definitions for use with the UTF-8 encoding. It
4 * actually also works with the variant UTF-8 encoding called UTF-EBCDIC, and
5 * hides almost all of the differences between these from the caller. In other
6 * words, someone should #include this file, and if the code is being compiled
7 * on an EBCDIC platform, things should mostly just work.
8 *
2eee27d7
SS
9 * Copyright (C) 2000, 2001, 2002, 2005, 2006, 2007, 2009,
10 * 2010, 2011 by Larry Wall and others
a0ed51b3
LW
11 *
12 * You may distribute under the terms of either the GNU General Public
13 * License or the Artistic License, as specified in the README file.
14 *
15 */
16
57f0e7e2
KW
17#ifndef H_UTF8 /* Guard against recursive inclusion */
18#define H_UTF8 1
19
39e02b42 20/* Use UTF-8 as the default script encoding?
1e54db1a 21 * Turning this on will break scripts having non-UTF-8 binary
39e02b42
JH
22 * data (such as Latin-1) in string literals. */
23#ifdef USE_UTF8_SCRIPTS
24# define USE_UTF8_IN_NAMES (!IN_BYTES)
25#else
26# define USE_UTF8_IN_NAMES (PL_hints & HINT_UTF8)
27#endif
28
051a06d4
KW
29/* For to_utf8_fold_flags, q.v. */
30#define FOLD_FLAGS_LOCALE 0x1
31#define FOLD_FLAGS_FULL 0x2
a0270393 32#define FOLD_FLAGS_NOMIX_ASCII 0x4
051a06d4 33
83199d38
KW
34/* For _core_swash_init(), internal core use only */
35#define _CORE_SWASH_INIT_USER_DEFINED_PROPERTY 0x1
5d3d13d1 36#define _CORE_SWASH_INIT_RETURN_IF_UNDEF 0x2
87367d5f 37#define _CORE_SWASH_INIT_ACCEPT_INVLIST 0x4
83199d38 38
a0270393 39#define to_uni_fold(c, p, lenp) _to_uni_fold_flags(c, p, lenp, FOLD_FLAGS_FULL)
051a06d4
KW
40#define to_utf8_fold(c, p, lenp) _to_utf8_fold_flags(c, p, lenp, \
41 FOLD_FLAGS_FULL, NULL)
42#define to_utf8_lower(a,b,c) _to_utf8_lower_flags(a,b,c,0, NULL)
43#define to_utf8_upper(a,b,c) _to_utf8_upper_flags(a,b,c,0, NULL)
44#define to_utf8_title(a,b,c) _to_utf8_title_flags(a,b,c,0, NULL)
36bb2ab6 45
fd7cb289
RGS
46/* Source backward compatibility. */
47#define uvuni_to_utf8(d, uv) uvuni_to_utf8_flags(d, uv, 0)
48#define is_utf8_string_loc(s, len, ep) is_utf8_string_loclen(s, len, ep, 0)
49
eda9cac1
KW
50#define foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
51 foldEQ_utf8_flags(s1, pe1, l1, u1, s2, pe2, l2, u2, 0)
a33c29bc 52#define FOLDEQ_UTF8_NOMIX_ASCII (1 << 0)
5e64d0fa 53#define FOLDEQ_UTF8_LOCALE (1 << 1)
18f762c3
KW
54#define FOLDEQ_S1_ALREADY_FOLDED (1 << 2)
55#define FOLDEQ_S2_ALREADY_FOLDED (1 << 3)
a33c29bc 56
e6226b18
KW
57/*
58=for apidoc ibcmp_utf8
59
60This is a synonym for (! foldEQ_utf8())
61
62=cut
63*/
64#define ibcmp_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2) \
65 cBOOL(! foldEQ_utf8(s1, pe1, l1, u1, s2, pe2, l2, u2))
66
1d72bdf6
NIS
67#ifdef EBCDIC
68/* The equivalent of these macros but implementing UTF-EBCDIC
69 are in the following header file:
70 */
71
72#include "utfebcdic.h"
fd7cb289 73
d06134e5 74#else /* ! EBCDIC */
73c4f7a1
GS
75START_EXTERN_C
76
a0ed51b3 77#ifdef DOINIT
6f06b55f 78EXTCONST unsigned char PL_utf8skip[] = {
b2635aa8
KW
79/* 0x00 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
80/* 0x10 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
81/* 0x20 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
82/* 0x30 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
83/* 0x40 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
84/* 0x50 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
85/* 0x60 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
86/* 0x70 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* ascii */
87/* 0x80 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
88/* 0x90 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
89/* 0xA0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
90/* 0xB0 */ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, /* bogus: continuation byte */
91/* 0xC0 */ 2,2, /* overlong */
92/* 0xC2 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0080 to U+03FF */
93/* 0xD0 */ 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, /* U+0400 to U+07FF */
94/* 0xE0 */ 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, /* U+0800 to U+FFFF */
95/* 0xF0 */ 4,4,4,4,4,4,4,4,5,5,5,5,6,6, /* above BMP to 2**31 - 1 */
96/* 0xFE */ 7,13, /* Perl extended (never was official UTF-8). Up to 72bit
97 allowed (64-bit + reserved). */
a0ed51b3
LW
98};
99#else
6f06b55f 100EXTCONST unsigned char PL_utf8skip[];
a0ed51b3
LW
101#endif
102
73c4f7a1 103END_EXTERN_C
7e2040f0 104
b96a92fb 105#include "regcharclass.h"
ec34087a
KW
106#include "unicode_constants.h"
107
1d72bdf6
NIS
108/* Native character to iso-8859-1 */
109#define NATIVE_TO_ASCII(ch) (ch)
110#define ASCII_TO_NATIVE(ch) (ch)
111/* Transform after encoding */
112#define NATIVE_TO_UTF(ch) (ch)
428921e5 113#define NATIVE_TO_I8(ch) NATIVE_TO_UTF(ch) /* a clearer synonym */
1d72bdf6 114#define UTF_TO_NATIVE(ch) (ch)
428921e5 115#define I8_TO_NATIVE(ch) UTF_TO_NATIVE(ch)
1d72bdf6
NIS
116/* Transforms in wide UV chars */
117#define UNI_TO_NATIVE(ch) (ch)
118#define NATIVE_TO_UNI(ch) (ch)
119/* Transforms in invariant space */
120#define NATIVE_TO_NEED(enc,ch) (ch)
121#define ASCII_TO_NEED(enc,ch) (ch)
d7578b48 122
d06134e5 123/* As there are no translations, avoid the function wrapper */
1754c1a1 124#define utf8n_to_uvchr utf8n_to_uvuni
6dd9dce9 125#define valid_utf8_to_uvchr valid_utf8_to_uvuni
1754c1a1 126#define uvchr_to_utf8 uvuni_to_utf8
2b9d42f0 127
877d9f0d 128/*
9041c2e3 129
8c007b5a 130 The following table is from Unicode 3.2.
877d9f0d
JH
131
132 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
133
375122d7 134 U+0000..U+007F 00..7F
e1b711da 135 U+0080..U+07FF * C2..DF 80..BF
37e2e78e 136 U+0800..U+0FFF E0 * A0..BF 80..BF
375122d7 137 U+1000..U+CFFF E1..EC 80..BF 80..BF
e1b711da 138 U+D000..U+D7FF ED 80..9F 80..BF
37e2e78e 139 U+D800..U+DFFF +++++++ utf16 surrogates, not legal utf8 +++++++
375122d7 140 U+E000..U+FFFF EE..EF 80..BF 80..BF
37e2e78e 141 U+10000..U+3FFFF F0 * 90..BF 80..BF 80..BF
877d9f0d
JH
142 U+40000..U+FFFFF F1..F3 80..BF 80..BF 80..BF
143 U+100000..U+10FFFF F4 80..8F 80..BF 80..BF
b2635aa8
KW
144 Below are non-Unicode code points
145 U+110000..U+13FFFF F4 90..BF 80..BF 80..BF
146 U+110000..U+1FFFFF F5..F7 80..BF 80..BF 80..BF
147 U+200000: F8.. * 88..BF 80..BF 80..BF 80..BF
877d9f0d 148
e1b711da 149Note the gaps before several of the byte entries above marked by '*'. These are
37e2e78e
KW
150caused by legal UTF-8 avoiding non-shortest encodings: it is technically
151possible to UTF-8-encode a single code point in different ways, but that is
152explicitly forbidden, and the shortest possible encoding should always be used
15824458 153(and that is what Perl does). The non-shortest ones are called 'overlongs'.
8c007b5a 154
877d9f0d
JH
155 */
156
8c007b5a
JH
157/*
158 Another way to look at it, as bits:
159
b2635aa8 160 Code Points 1st Byte 2nd Byte 3rd Byte 4th Byte
8c007b5a 161
b2635aa8
KW
162 0aaa aaaa 0aaa aaaa
163 0000 0bbb bbaa aaaa 110b bbbb 10aa aaaa
164 cccc bbbb bbaa aaaa 1110 cccc 10bb bbbb 10aa aaaa
165 00 000d ddcc cccc bbbb bbaa aaaa 1111 0ddd 10cc cccc 10bb bbbb 10aa aaaa
8c007b5a
JH
166
167As you can see, the continuation bytes all begin with C<10>, and the
e1b711da 168leading bits of the start byte tell how many bytes there are in the
8c007b5a
JH
169encoded character.
170
65ab9279
TC
171Perl's extended UTF-8 means we can have start bytes up to FF.
172
8c007b5a
JH
173*/
174
15824458
KW
175/* Is the representation of the Unicode code point 'c' the same regardless of
176 * being encoded in UTF-8 or not? */
c4d5f83a 177#define UNI_IS_INVARIANT(c) (((UV)c) < 0x80)
15824458
KW
178
179/* Is the UTF8-encoded byte 'c' part of a variant sequence in UTF-8? This is
180 * the inverse of UTF8_IS_INVARIANT */
181#define UTF8_IS_CONTINUED(c) (((U8)c) & 0x80)
182
183/* Is the byte 'c' the first byte of a multi-byte UTF8-8 encoded sequence?
184 * This doesn't catch invariants (they are single-byte). It also excludes the
185 * illegal overlong sequences that begin with C0 and C1. */
0447e8df 186#define UTF8_IS_START(c) (((U8)c) >= 0xc2)
15824458
KW
187
188/* Is the byte 'c' part of a multi-byte UTF8-8 encoded sequence, and not the
189 * first byte thereof? */
e021c6e6 190#define UTF8_IS_CONTINUATION(c) ((((U8)c) & 0xC0) == 0x80)
0ae1fa71 191
15824458
KW
192/* Is the UTF8-encoded byte 'c' the first byte of a two byte sequence? Use
193 * UTF8_IS_NEXT_CHAR_DOWNGRADEABLE() instead if the input isn't known to
194 * be well-formed. Masking with 0xfe allows the low bit to be 0 or 1; thus
195 * this matches 0xc[23]. */
559c7f10 196#define UTF8_IS_DOWNGRADEABLE_START(c) (((U8)(c) & 0xfe) == 0xc2)
4ab10950 197
15824458
KW
198/* Is the UTF8-encoded byte 'c' the first byte of a sequence of bytes that
199 * represent a code point > 255? */
7b4252f4 200#define UTF8_IS_ABOVE_LATIN1(c) ((U8)(c) >= 0xc4)
8850bf83 201
15824458
KW
202/* This defines the 1-bits that are to be in the first byte of a multi-byte
203 * UTF-8 encoded character that give the number of bytes that comprise the
204 * character.
205 * */
22901f30 206#define UTF_START_MARK(len) (((len) > 7) ? 0xFF : (0xFE << (7-(len))))
b2635aa8
KW
207
208/* Masks out the initial one bits in a start byte, leaving the real data ones.
209 * Doesn't work on an invariant byte */
22901f30 210#define UTF_START_MASK(len) (((len) >= 7) ? 0x00 : (0x1F >> ((len)-2)))
1d72bdf6 211
15824458
KW
212/* This defines the bits that are to be in the continuation bytes of a multi-byte
213 * UTF-8 encoded character that indicate it is a continuation byte. */
1d72bdf6 214#define UTF_CONTINUATION_MARK 0x80
15824458
KW
215
216/* This is the number of low-order bits a continuation byte in a UTF-8 encoded
217 * sequence contributes to the specification of the code point. In the bit
218 * maps above, you see that the first 2 bits are a constant '10', leaving 6 of
219 * real information */
1d72bdf6 220#define UTF_ACCUMULATION_SHIFT 6
b2635aa8
KW
221
222/* 2**UTF_ACCUMULATION_SHIFT - 1 */
1d72bdf6 223#define UTF_CONTINUATION_MASK ((U8)0x3f)
c512ce4f 224
15824458
KW
225/* If a value is anded with this, and the result is non-zero, then using the
226 * original value in UTF8_ACCUMULATE will overflow, shifting bits off the left
227 * */
eb83ed87 228#define UTF_ACCUMULATION_OVERFLOW_MASK \
bb88be5f
KW
229 (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) \
230 - UTF_ACCUMULATION_SHIFT))
eb83ed87 231
1d68d6cd 232#ifdef HAS_QUAD
5bbb0b5a 233#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
234 (uv) < 0x800 ? 2 : \
235 (uv) < 0x10000 ? 3 : \
236 (uv) < 0x200000 ? 4 : \
237 (uv) < 0x4000000 ? 5 : \
238 (uv) < 0x80000000 ? 6 : \
9041c2e3 239 (uv) < UTF8_QUAD_MAX ? 7 : 13 )
1d68d6cd
SC
240#else
241/* No, I'm not even going to *TRY* putting #ifdef inside a #define */
5bbb0b5a 242#define UNISKIP(uv) ( (uv) < 0x80 ? 1 : \
1d68d6cd
SC
243 (uv) < 0x800 ? 2 : \
244 (uv) < 0x10000 ? 3 : \
245 (uv) < 0x200000 ? 4 : \
246 (uv) < 0x4000000 ? 5 : \
247 (uv) < 0x80000000 ? 6 : 7 )
248#endif
249
03c76984
KW
250/* How wide can a single UTF-8 encoded character become in bytes. */
251/* NOTE: Strictly speaking Perl's UTF-8 should not be called UTF-8 since UTF-8
252 * is an encoding of Unicode, and Unicode's upper limit, 0x10FFFF, can be
253 * expressed with 4 bytes. However, Perl thinks of UTF-8 as a way to encode
254 * non-negative integers in a binary format, even those above Unicode */
255#define UTF8_MAXBYTES 13
256
257/* The maximum number of UTF-8 bytes a single Unicode character can
258 * uppercase/lowercase/fold into. Unicode guarantees that the maximum
259 * expansion is 3 characters. On ASCIIish platforms, the highest Unicode
260 * character occupies 4 bytes, therefore this number would be 12, but this is
261 * smaller than the maximum width a single above-Unicode character can occupy,
262 * so use that instead */
263#if UTF8_MAXBYTES < 12
264#error UTF8_MAXBYTES must be at least 12
265#endif
266
267#define UTF8_MAXBYTES_CASE UTF8_MAXBYTES
268
d06134e5
KW
269#endif /* EBCDIC vs ASCII */
270
271/* Rest of these are attributes of Unicode and perl's internals rather than the
272 * encoding, or happen to be the same in both ASCII and EBCDIC (at least at
273 * this level; the macros that some of these call may have different
274 * definitions in the two encodings */
275
276#define NATIVE8_TO_UNI(ch) NATIVE_TO_ASCII(ch) /* a clearer synonym */
277
bb88be5f
KW
278#define UTF8_ACCUMULATE(old, new) (((old) << UTF_ACCUMULATION_SHIFT) \
279 | (((U8)new) & UTF_CONTINUATION_MASK))
d06134e5 280
4ab10950
KW
281/* This works in the face of malformed UTF-8. */
282#define UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(s, e) (UTF8_IS_DOWNGRADEABLE_START(*s) \
283 && ( (e) - (s) > 1) \
284 && UTF8_IS_CONTINUATION(*((s)+1)))
285
2950f2a7
KW
286/* Convert a two (not one) byte utf8 character to a unicode code point value.
287 * Needs just one iteration of accumulate. Should not be used unless it is
288 * known that the two bytes are legal: 1) two-byte start, and 2) continuation.
289 * Note that the result can be larger than 255 if the input character is not
290 * downgradable */
291#define TWO_BYTE_UTF8_TO_UNI(HI, LO) \
292 UTF8_ACCUMULATE((NATIVE_TO_UTF(HI) & UTF_START_MASK(2)), \
293 NATIVE_TO_UTF(LO))
294
15824458
KW
295/* How many bytes in the UTF-8 encoded character whose first (perhaps only)
296 * byte is pointed to by 's' */
d06134e5
KW
297#define UTF8SKIP(s) PL_utf8skip[*(const U8*)(s)]
298
15824458
KW
299/* Is the byte 'c' the same character when encoded in UTF-8 as when not. This
300 * works on both UTF-8 encoded strings and non-encoded, as it returns TRUE in
301 * each for the exact same set of bit patterns. (And it works on any byte in a
302 * UTF-8 encoded string) */
d06134e5 303#define UTF8_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE_TO_UTF(c))
15824458 304
d06134e5
KW
305#define NATIVE_IS_INVARIANT(c) UNI_IS_INVARIANT(NATIVE8_TO_UNI(c))
306
307#define MAX_PORTABLE_UTF8_TWO_BYTE 0x3FF /* constrained by EBCDIC */
308
309/* The macros in the next sets are used to generate the two utf8 or utfebcdic
310 * bytes from an ordinal that is known to fit into two bytes; it must be less
311 * than 0x3FF to work across both encodings. */
312/* Nocast allows these to be used in the case label of a switch statement */
428921e5
KW
313#define UTF8_TWO_BYTE_HI_nocast(c) NATIVE_TO_I8(((c) \
314 >> UTF_ACCUMULATION_SHIFT) | (0xFF & UTF_START_MARK(2)))
315#define UTF8_TWO_BYTE_LO_nocast(c) NATIVE_TO_I8(((c) & UTF_CONTINUATION_MASK) \
316 | UTF_CONTINUATION_MARK)
d06134e5
KW
317
318#define UTF8_TWO_BYTE_HI(c) ((U8) (UTF8_TWO_BYTE_HI_nocast(c)))
319#define UTF8_TWO_BYTE_LO(c) ((U8) (UTF8_TWO_BYTE_LO_nocast(c)))
320
321/* This name is used when the source is a single byte */
322#define UTF8_EIGHT_BIT_HI(c) UTF8_TWO_BYTE_HI((U8)(c))
323#define UTF8_EIGHT_BIT_LO(c) UTF8_TWO_BYTE_LO((U8)(c))
324
7e2040f0 325/*
e3036cf4 326 * 'UTF' is whether or not p is encoded in UTF8. The names 'foo_lazy_if' stem
20df05f4
KW
327 * from an earlier version of these macros in which they didn't call the
328 * foo_utf8() macros (i.e. were 'lazy') unless they decided that *p is the
329 * beginning of a utf8 character. Now that foo_utf8() determines that itself,
330 * no need to do it again here
7e2040f0 331 */
e3036cf4
KW
332#define isIDFIRST_lazy_if(p,UTF) ((IN_BYTES || !UTF ) \
333 ? isIDFIRST(*(p)) \
334 : isIDFIRST_utf8((const U8*)p))
32636478
KW
335#define isWORDCHAR_lazy_if(p,UTF) ((IN_BYTES || (!UTF )) \
336 ? isWORDCHAR(*(p)) \
337 : isWORDCHAR_utf8((const U8*)p))
338#define isALNUM_lazy_if(p,UTF) isWORDCHAR_lazy_if(p,UTF)
1d72bdf6 339
89ebb4a3
JH
340#define UTF8_MAXLEN UTF8_MAXBYTES
341
8cb75cc8
KW
342/* A Unicode character can fold to up to 3 characters */
343#define UTF8_MAX_FOLD_CHAR_EXPAND 3
344
a98fe34d 345#define IN_BYTES (CopHINTS_get(PL_curcop) & HINT_BYTES)
0064a8a9 346#define DO_UTF8(sv) (SvUTF8(sv) && !IN_BYTES)
66cbab2c 347#define IN_UNI_8_BIT \
b36bf33f
KW
348 (CopHINTS_get(PL_curcop) & (HINT_UNI_8_BIT|HINT_LOCALE_NOT_CHARS) \
349 && ! IN_LOCALE_RUNTIME && ! IN_BYTES)
350
1d72bdf6 351
c76687c5
KW
352#define UTF8_ALLOW_EMPTY 0x0001 /* Allow a zero length string */
353
354/* Allow first byte to be a continuation byte */
1d72bdf6 355#define UTF8_ALLOW_CONTINUATION 0x0002
c76687c5
KW
356
357/* Allow second... bytes to be non-continuation bytes */
1d72bdf6 358#define UTF8_ALLOW_NON_CONTINUATION 0x0004
949cf498
KW
359
360/* expecting more bytes than were available in the string */
361#define UTF8_ALLOW_SHORT 0x0008
362
363/* Overlong sequence; i.e., the code point can be specified in fewer bytes. */
364#define UTF8_ALLOW_LONG 0x0010
365
366#define UTF8_DISALLOW_SURROGATE 0x0020 /* Unicode surrogates */
367#define UTF8_WARN_SURROGATE 0x0040
368
369#define UTF8_DISALLOW_NONCHAR 0x0080 /* Unicode non-character */
370#define UTF8_WARN_NONCHAR 0x0100 /* code points */
371
372#define UTF8_DISALLOW_SUPER 0x0200 /* Super-set of Unicode: code */
373#define UTF8_WARN_SUPER 0x0400 /* points above the legal max */
374
375/* Code points which never were part of the original UTF-8 standard, the first
376 * byte of which is a FE or FF on ASCII platforms. */
377#define UTF8_DISALLOW_FE_FF 0x0800
378#define UTF8_WARN_FE_FF 0x1000
379
380#define UTF8_CHECK_ONLY 0x2000
381
382/* For backwards source compatibility. They do nothing, as the default now
383 * includes what they used to mean. The first one's meaning was to allow the
384 * just the single non-character 0xFFFF */
385#define UTF8_ALLOW_FFFF 0
386#define UTF8_ALLOW_SURROGATE 0
387
33d9abfb 388#define UTF8_DISALLOW_ILLEGAL_INTERCHANGE (UTF8_DISALLOW_SUPER|UTF8_DISALLOW_NONCHAR|UTF8_DISALLOW_SURROGATE|UTF8_DISALLOW_FE_FF)
949cf498 389#define UTF8_WARN_ILLEGAL_INTERCHANGE \
33d9abfb 390 (UTF8_WARN_SUPER|UTF8_WARN_NONCHAR|UTF8_WARN_SURROGATE|UTF8_WARN_FE_FF)
949cf498
KW
391#define UTF8_ALLOW_ANY \
392 (~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
393#define UTF8_ALLOW_ANYUV \
394 (UTF8_ALLOW_EMPTY \
395 & ~(UTF8_DISALLOW_ILLEGAL_INTERCHANGE|UTF8_WARN_ILLEGAL_INTERCHANGE))
9f7f3913
ST
396#define UTF8_ALLOW_DEFAULT (ckWARN(WARN_UTF8) ? 0 : \
397 UTF8_ALLOW_ANYUV)
1d72bdf6 398
7131f24d
KW
399/* Surrogates, non-character code points and above-Unicode code points are
400 * problematic in some contexts. This allows code that needs to check for
401 * those to to quickly exclude the vast majority of code points it will
402 * encounter */
b96a92fb
KW
403#define UTF8_FIRST_PROBLEMATIC_CODE_POINT_FIRST_BYTE \
404 FIRST_SURROGATE_UTF8_FIRST_BYTE
7131f24d 405
b96a92fb
KW
406#define UTF8_IS_SURROGATE(s) cBOOL(is_SURROGATE_utf8(s))
407#define UTF8_IS_REPLACEMENT(s, send) cBOOL(is_REPLACEMENT_utf8_safe(s,send))
7131f24d
KW
408
409/* ASCII EBCDIC I8
410 * U+10FFFF: \xF4\x8F\xBF\xBF \xF9\xA1\xBF\xBF\xBF max legal Unicode
411 * U+110000: \xF4\x90\x80\x80 \xF9\xA2\xA0\xA0\xA0
412 * U+110001: \xF4\x90\x80\x81 \xF9\xA2\xA0\xA0\xA1
413 */
414#ifdef EBCDIC /* Both versions assume well-formed UTF8 */
0404c92a
KW
415# define UTF8_IS_SUPER(s) (NATIVE_TO_I8(* (U8*) (s)) >= 0xF9 \
416 && (NATIVE_TO_I8(* (U8*) (s)) > 0xF9 \
417 || (NATIVE_TO_I8(* (U8*) ((s)) + 1 >= 0xA2))))
7131f24d 418#else
0404c92a
KW
419# define UTF8_IS_SUPER(s) (*(U8*) (s) >= 0xF4 \
420 && (*(U8*) (s) > 0xF4 || (*((U8*) (s) + 1) >= 0x90)))
7131f24d
KW
421#endif
422
b96a92fb
KW
423/* These are now machine generated, and the 'given' clause is no longer
424 * applicable */
425#define UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s) \
426 cBOOL(is_NONCHAR_utf8(s))
427#define UTF8_IS_NONCHAR_(s) \
428 UTF8_IS_NONCHAR_GIVEN_THAT_NON_SUPER_AND_GE_PROBLEMATIC(s)
7131f24d 429
c867b360
JH
430#define UNICODE_SURROGATE_FIRST 0xD800
431#define UNICODE_SURROGATE_LAST 0xDFFF
432#define UNICODE_REPLACEMENT 0xFFFD
433#define UNICODE_BYTE_ORDER_MARK 0xFEFF
1d72bdf6 434
b851fbc1 435/* Though our UTF-8 encoding can go beyond this,
c76687c5 436 * let's be conservative and do as Unicode says. */
b851fbc1
JH
437#define PERL_UNICODE_MAX 0x10FFFF
438
949cf498
KW
439#define UNICODE_WARN_SURROGATE 0x0001 /* UTF-16 surrogates */
440#define UNICODE_WARN_NONCHAR 0x0002 /* Non-char code points */
441#define UNICODE_WARN_SUPER 0x0004 /* Above 0x10FFFF */
442#define UNICODE_WARN_FE_FF 0x0008 /* Above 0x10FFFF */
443#define UNICODE_DISALLOW_SURROGATE 0x0010
444#define UNICODE_DISALLOW_NONCHAR 0x0020
445#define UNICODE_DISALLOW_SUPER 0x0040
446#define UNICODE_DISALLOW_FE_FF 0x0080
bb88be5f
KW
447#define UNICODE_WARN_ILLEGAL_INTERCHANGE \
448 (UNICODE_WARN_SURROGATE|UNICODE_WARN_NONCHAR|UNICODE_WARN_SUPER)
449#define UNICODE_DISALLOW_ILLEGAL_INTERCHANGE \
450 (UNICODE_DISALLOW_SURROGATE|UNICODE_DISALLOW_NONCHAR|UNICODE_DISALLOW_SUPER)
949cf498
KW
451
452/* For backward source compatibility, as are now the default */
453#define UNICODE_ALLOW_SURROGATE 0
454#define UNICODE_ALLOW_SUPER 0
455#define UNICODE_ALLOW_ANY 0
b851fbc1 456
1d72bdf6
NIS
457#define UNICODE_IS_SURROGATE(c) ((c) >= UNICODE_SURROGATE_FIRST && \
458 (c) <= UNICODE_SURROGATE_LAST)
a10ec373 459#define UNICODE_IS_REPLACEMENT(c) ((c) == UNICODE_REPLACEMENT)
872c91ae 460#define UNICODE_IS_BYTE_ORDER_MARK(c) ((c) == UNICODE_BYTE_ORDER_MARK)
7131f24d
KW
461#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) \
462 /* The other noncharacters end in FFFE or FFFF, which \
463 * the mask below catches both of, but beyond the last \
464 * official unicode code point, they aren't \
465 * noncharacters, since those aren't Unicode \
466 * characters at all */ \
467 || ((((c & 0xFFFE) == 0xFFFE)) && ! UNICODE_IS_SUPER(c)))
468#define UNICODE_IS_SUPER(c) ((c) > PERL_UNICODE_MAX)
469#define UNICODE_IS_FE_FF(c) ((c) > 0x7FFFFFFF)
1d72bdf6 470
f067b878
NA
471#ifdef HAS_QUAD
472# define UTF8_QUAD_MAX UINT64_C(0x1000000000)
473#endif
3bd709b1 474
ec34087a
KW
475#define LATIN_SMALL_LETTER_SHARP_S LATIN_SMALL_LETTER_SHARP_S_NATIVE
476#define LATIN_SMALL_LETTER_Y_WITH_DIAERESIS \
477 LATIN_SMALL_LETTER_Y_WITH_DIAERESIS_NATIVE
478#define MICRO_SIGN MICRO_SIGN_NATIVE
479#define LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE \
480 LATIN_CAPITAL_LETTER_A_WITH_RING_ABOVE_NATIVE
481#define LATIN_SMALL_LETTER_A_WITH_RING_ABOVE \
482 LATIN_SMALL_LETTER_A_WITH_RING_ABOVE_NATIVE
09091399
JH
483#define UNICODE_GREEK_CAPITAL_LETTER_SIGMA 0x03A3
484#define UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA 0x03C2
485#define UNICODE_GREEK_SMALL_LETTER_SIGMA 0x03C3
9dcbe121 486#define GREEK_SMALL_LETTER_MU 0x03BC
78a0d3cc
KW
487#define GREEK_CAPITAL_LETTER_MU 0x039C /* Upper and title case of MICRON */
488#define LATIN_CAPITAL_LETTER_Y_WITH_DIAERESIS 0x0178 /* Also is title case */
97298f37 489#define LATIN_CAPITAL_LETTER_SHARP_S 0x1E9E
8d64d87f
KW
490#define LATIN_SMALL_LETTER_LONG_S 0x017F
491#define KELVIN_SIGN 0x212A
492#define ANGSTROM_SIGN 0x212B
09091399 493
9e55ce06 494#define UNI_DISPLAY_ISPRINT 0x0001
c728cb41
JH
495#define UNI_DISPLAY_BACKSLASH 0x0002
496#define UNI_DISPLAY_QQ (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
497#define UNI_DISPLAY_REGEX (UNI_DISPLAY_ISPRINT|UNI_DISPLAY_BACKSLASH)
9e55ce06 498
5cd46e1f
KW
499#define ANYOF_FOLD_SHARP_S(node, input, end) \
500 (ANYOF_BITMAP_TEST(node, LATIN_SMALL_LETTER_SHARP_S) && \
137165a6 501 (ANYOF_NONBITMAP(node)) && \
39065660 502 (ANYOF_FLAGS(node) & ANYOF_LOC_NONBITMAP_FOLD) && \
07b6858f
JH
503 ((end) > (input) + 1) && \
504 toLOWER((input)[0]) == 's' && \
505 toLOWER((input)[1]) == 's')
ebc501f0 506#define SHARP_S_SKIP 2
3b0fc154 507
a4f7a67c
KW
508/* If you want to exclude surrogates, and beyond legal Unicode, see the blame
509 * log for earlier versions which gave details for these */
4d646140
KW
510
511#ifndef EBCDIC
512/* This was generated by regen/regcharclass.pl, and then moved here. The lines
513 * that generated it were then commented out. This was done solely because it
514 * takes on the order of 10 minutes to generate, and is never going to change.
515 * The EBCDIC equivalent hasn't been commented out in regcharclass.pl, so it
516 * should generate and run the correct stuff */
39a0f513
YO
517/*
518 UTF8_CHAR: Matches utf8 from 1 to 4 bytes
519
520 0x0 - 0x1FFFFF
521*/
4d646140
KW
522/*** GENERATED CODE ***/
523#define is_UTF8_CHAR_utf8_safe(s,e) \
524( ((e)-(s) > 3) ? \
525 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
526 : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
527 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
528 : ( 0xE0 == ((U8*)s)[0] ) ? \
529 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
530 : ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) ? \
531 ( ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
532 : ( 0xF0 == ((U8*)s)[0] ) ? \
533 ( ( ( ( 0x90 <= ((U8*)s)[1] && ((U8*)s)[1] <= 0xBF ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
39a0f513 534 : ( ( ( ( 0xF1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xF7 ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[3] & 0xC0 ) == 0x80 ) ) ? 4 : 0 )\
4d646140
KW
535: ((e)-(s) > 2) ? \
536 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
537 : ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) ? \
538 ( ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ? 2 : 0 ) \
539 : ( 0xE0 == ((U8*)s)[0] ) ? \
540 ( ( ( ( ((U8*)s)[1] & 0xE0 ) == 0xA0 ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
39a0f513 541 : ( ( ( 0xE1 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xEF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) && ( ( ((U8*)s)[2] & 0xC0 ) == 0x80 ) ) ? 3 : 0 )\
4d646140
KW
542: ((e)-(s) > 1) ? \
543 ( ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) ? 1 \
39a0f513 544 : ( ( 0xC2 <= ((U8*)s)[0] && ((U8*)s)[0] <= 0xDF ) && ( ( ((U8*)s)[1] & 0xC0 ) == 0x80 ) ) ? 2 : 0 )\
4d646140
KW
545: ((e)-(s) > 0) ? \
546 ( ( ((U8*)s)[0] & 0x80 ) == 0x00 ) \
547: 0 )
548#endif
3b0fc154
JH
549
550/* IS_UTF8_CHAR(p) is strictly speaking wrong (not UTF-8) because it
551 * (1) allows UTF-8 encoded UTF-16 surrogates
552 * (2) it allows code points past U+10FFFF.
553 * The Perl_is_utf8_char() full "slow" code will handle the Perl
554 * "extended UTF-8". */
4d646140 555#define IS_UTF8_CHAR(p, n) (is_UTF8_CHAR_utf8_safe(p, (p) + (n)) == n)
2f454f11 556
4d646140
KW
557/* regen/regcharclass.pl generates is_UTF8_CHAR_utf8_safe() macros for up to
558 * these number of bytes. So this has to be coordinated with it */
559#ifdef EBCDIC
2f454f11 560# define IS_UTF8_CHAR_FAST(n) ((n) <= 5)
4d646140
KW
561#else
562# define IS_UTF8_CHAR_FAST(n) ((n) <= 4)
563#endif
e9a8c099 564
57f0e7e2
KW
565#endif /* H_UTF8 */
566
e9a8c099
MHM
567/*
568 * Local variables:
569 * c-indentation-style: bsd
570 * c-basic-offset: 4
14d04a33 571 * indent-tabs-mode: nil
e9a8c099
MHM
572 * End:
573 *
14d04a33 574 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 575 */